Architecture with One or More FPGAsThe first architecture, shown in Figure 1, uses one or more field-programmable gate arrays (FPGAs) or application-specific integrated circuits (ASICs) to receive and transmit antenna data typically through common public radio interface (CPRI) or the Open Base Station Architecture Initiative (OBSAI), a common antenna interface standard. This is used when the physical layer protocol (PHY) layer of the wireless protocol is implemented in the FPGA/ASIC.
click to enlarge
Figure 1. Serial RapidIO Wireless Basestation Using an FPGA/ASIC.
When using an FPGA in this architecture, the base station is completely software programmable and scalable. It can be described as a “software configurable base station.” System designers could increase or decrease the size of the FPGA and add or remove DSPs to get a smaller or bigger base station that could service more or less subscribers. Also, for the MIMO algorithm, designers could also group users in smaller or bigger groups, which would require more or less processing power from the FPGA and DSPs.
Architecture Without FPGA or ASIC
click to enlarge
Figure 2. DSP-only Serial RapidIO Wireless Basestation.
Similarly, a lot of control traffic exists between the control processor and the DSP for QOS and management functions for each user. Serial RapidIO is ideal in architectures that utilize logical layer functions, such as direct memory transactions and/or RapidIO Messaging, which is an effective way to send a large message for up to 4Kbs. The RapidIO Messaging is a push architecture and enjoys the lowest system latency over other protocols. The Serial RapidIO switch latency is also minimal at around 100ns, creating a seamless DSP-to-DSP transfer. This architecture is also scalable and can also be described as a software configurable base station.
In the architecture illustrated in Figure 2, not only can designers easily define smaller or bigger base stations that can service more or less users, but they can also target different wireless standards with the same architecture. Only the software needs to change to make the architecture a common platform between standards.
For the architectures in Figure 1 and Figure 2, the S-RIO data flows are vastly different. Similarly, the same architecture that would target different wireless standards, such as WCDMA vs. LTE, would require a different S-RIO data flow. For that reason, it would be wise to model the S-RIO traffic scenarios within the system to ensure the size and speed of each link is adequate to perform the necessary algorithm within the timing budget of each standard.
S-RIO offers link widths of 1x, 2x and 4x, and link speeds ranging from 1.25 to 6.25 Gb/s. Figure 1 and Figure 2 show S-RIO links of 4x, but some systems can be performed with smaller links where appropriate depending on the number of users serviced and the capabilities of each DSP involved.
ConclusionSerial RapidIO is the only interconnect technology defined to address a distributed array of processors arranged in any topology with easy routing and adaptability. Its simplistic approach to routing packets based on the Destination ID field makes it easy for software implementers to discover new nodes in a system or redistribute traffic in the case of processor failures.
Serial RapidIO is scalable while maintaining the same software programming model without complications of memory map management common with other protocols. No other protocol can make the two architectures discussed above into “Software Programmable Base Stations.” For more information about the Serial RapidIO protocol, please visit www.rapidio.org and for more information about IDT Serial RapidIO switches and enablement tools, please visit http://www.IDT.com/go/SRIOSwitches.
Stephane Gagnon, director of product management, joined IDT in August 2000. For the past 10 years, Stephane has been involved in the RapidIO Trade Association Technical Working Group and currently holds the position of Chairman of the Trade Association Steering Committee.